Data for "The Dust Extinction Curve: Beyond R(V)"

24 million dust extinction curves, detemined from Gaia XP spectra, as described in Green, Zhang & Zhang (2025).

We represent the extinction curves using a set of 16 basis vectors. For each star, there are 16 coefficients, which can be used to reconstruct the extinction curves. After loading A_zp and G_subspace from the file G_subspace.json, and coeffs from the file coeffs.h5, the extinction curves can be reconstructed using:

    A = A_zp + np.sum(coeffs[None,:] * G_subspace[:,:], axis=1)

The output A will have shape (star, wavelength). The wavelengths at which A is sampled are stored in the field wavelengths_nm (in nanometers), in G_subspace.json.

The covariance matrix of the coefficients for each star is stored in the files coeffs_cov_?.h5. We store the diagonals and the upper triangles of the covariances separately. They can be reconstructed using the function reconstruct_symm_matrices from symm_matrix_utils.py:

    from symm_matrix_utils.py import reconstruct_symm_matrices    cov = reconstruct_symm_matrices(cov_diag, cov_triu_wo_diag)

Additionally, we store the inverse covariance matrices in the files coeffs_icov_?.h5, in the same manner as the covariance matrices.

The file source_info.h5 contains a few useful Gaia fields and parameter estimates (with corresponding uncertainties) from Zhang & Green (2025).

The file feature_EW.h5 contains the equivalent widths (in nanometers) of the VBS and the 770 and 850 nm extinction features.

Every file contains the Gaia DR3 source_id of every star, labeled gdr3_source_id.